Could not convert string to float - Error


#1

Hi, I am trying to use use AutoML on Desktop. I have my images, model and hyper parameters all set. I want to try in cloud version, but need to first figure out the issue on desktop before wasting $ on paid instance.

When I start training, I get the following error: could not convert string to float: './0.45/99788_17.tif'

the log shows:

raceback (most recent call last):
File “project\network.py”, line 608, in project.network.getOutputShapes
File “project\network.py”, line 565, in project.network.getLayerOutputShape
File “C:\Users\AnthonyResearch\AppData\Local\Programs\DeepLearningStudio\conda3\lib\site-packages\keras\layers\core.py”, line 476, in get_output_shape_for
'(got ’ + str(input_shape[1:]) + '. ’
ValueError: The shape of the input to “Flatten” is not fully defined (got (62, 0, 0). Make sure to pass a complete “input_shape” or “batch_input_shape” argument to the first layer in your model.

my CSV looks like this:

image

Is the problem the “period” character in my folder structure?


#2

Hi,
Can you send the snapshot of the model used and are you using more than one feature for output?
DLS supports multiple input but only one output right now.
Mail us small part of the dataset along with the csv file at support@deepcognition.ai
So that i can check the reason behind this and come out with a working model for your dataset.

Regards
Rajat